set_binary_path
Set the path to the binary file
LABSCO SUMMARY
IDA's analysis without an IDA window — the point is that it runs as a service.
The interactive-plugin route ties the agent to a session someone has to keep open. Running headless means the binary can be analyzed on a server, over sse, as part of something larger. The cost is that IDA Pro is still a licensed product you have to have.
What it is
A command-line MCP server that drives IDA Pro's headless mode over the headless-ida library, exposing analysis of a binary — its functions, variables and more — as tools.
What you get
- Binary analysis backed by IDA Pro itself, not a reimplementation of it
- Tools for inspecting and manipulating the functions and variables IDA has recovered
- Runs as a standalone CLI service rather than as an interactive IDA plugin
- Choice of transport — stdio for a local client, or sse for a networked one
Requirements
A licensed IDA Pro with headless support, and Python 3.12 or higher. Configuration goes in a .env file: IDA_PATH points at the idat executable, TRANSPORT selects sse or stdio, and HOST and PORT set the listen address when you use sse. Install with uv pip install -e . and start with uv run headless_ida_mcp_server. The tool implementations are adapted from ida-pro-mcp.
Setup effort
Build from source — git clone https://github.com/cnitlrt/headless-ida-mcp-server.git
USE CASES
01
Working through an unfamiliar binary
Let the assistant query IDA's recovered functions and variables and explain what it finds.
02
Analysing binaries on a machine with no desktop
Run it as a CLI service against idat rather than keeping an interactive IDA session open.
03
Reaching the analysis from another machine
Set TRANSPORT to sse with a host and port, and connect a client over the network.
04
Renaming and annotating as you go
Have the assistant apply changes to functions and variables in the database while it works.
21 tools
Listed in project docs. Your client may show a different list.
get_function
Get a function by address
get_function_by_name
Get a function by name
get_function_by_address
Get a function by address
get_current_address
Get the current address
get_current_function
Get the current function
convert_number
Convert a number to a different representation
list_functions
List all functions
decompile_checked
Decompile a function at the given address
decompile_function
Decompile a function at the given address
disassemble_function
Disassemble a function at the given address
get_xrefs_to
Get cross references to a given address
get_entry_points
Get all entry points of the binary
set_decompiler_comment
Set a comment for a given address in the function pseudocode
set_disassembly_comment
Set a comment for a given address in the function disassembly
refresh_decompiler_widget
Refresh the decompiler widget
refresh_decompiler_ctext
Refresh the decompiler ctext for a given function
rename_local_variable
Rename a local variable in a function
rename_function
Rename a function
set_function_prototype
Set a function's prototype
save_idb_file
Save the IDB file
BEFORE YOU INSTALL
What to know first
Tools you get
21 tools named in the project's documentation.
Requires
- Python 3.12 or higher
- IDA Pro with headless support (idat)
- uv
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
You clone this one yourself.
Server setup
TERMINAL
git clone https://github.com/cnitlrt/headless-ida-mcp-server.git
cd headless-ida-mcp-server
uv python install 3.12
uv venv --python 3.12
uv pip install -e .
cp .env_example .envIn .env, set IDA_PATH to IDA Pro's headless executable (idat), and TRANSPORT to stdio for a local client, or sse with HOST and PORT
MCP Clients
Any MCP client
ANY MCP CLIENT · JSON
{
"mcpServers": {
"ida": {
"command": "<YOUR_PATH_TO_UV>",
"args": [
"--directory",
"<YOUR_PATH_TO_HEADLESS_IDA_MCP_SERVER>",
"run",
"headless_ida_mcp_server"
]
}
}
}
